Fraud detection based on community change analysis using a machine learning model

Abstract

The disclosed embodiments include a method for performing financial fraud assessment that includes creating a machine learning model based on features used to identify financial fraud risk; receiving financial information associated with customer accounts; establishing communities for the customer accounts; creating a baseline set of the features for each of the communities; receiving new financial information associated with customer accounts; updating the communities for the customer accounts based on the new financial information; extracting an updated set of the features for each of the communities; and determining a difference between the baseline set of the features and the updated set of the features for each of the communities; and using the machine learning model to determine financial fraud risk for each of the communities based on the difference between the baseline set of the features and the updated set of the features for each of the communities.

What is claimed is: 
     
       1. A computer-implemented method for identifying financial fraud risk, the method comprising:
 receiving a training dataset that includes financial data that have been proven to be associated with financial fraud; 
 training a machine learning model based on an analysis of features from the training dataset to identify patterns in the financial data that have been proven to be associated with financial fraud; 
 receiving financial information associated with customer accounts; 
 establishing communities for the customer accounts; 
 creating a baseline set of the features for each of the communities, wherein the baseline set of the features comprises a first total dollar amount of all transactions between all the customer accounts in the community; 
 receiving new financial information associated with the customer accounts; 
 updating the communities for the customer accounts based on the new financial information; 
 extracting an updated set of the features for each of the communities, wherein the updated set of the features comprises a second total dollar amount of all transactions between all the customer accounts in the community; and 
 determining a difference between the baseline set of the features and the updated set of the features for each of the communities; 
 performing, by the machine learning model, a financial fraud risk assessment for each of the communities to identify one or more communities that have an increased risk of financial fraud by determining whether the difference between the baseline set of the features and the updated set of the features for each of the communities is indicative of an increased risk of fraud based on the patterns identified in the financial data that have been proven to be associated with financial fraud during training of the machine learning model, wherein there is an increased risk of financial fraud for the community when a difference in the second total dollar amount and the first total dollar amount for the community exceeds a threshold determined by the machine learning model that is indicative increased risk of financial fraud for the community; and 
 retraining the machine learning model based on an accuracy determination of the financial fraud risk assessment performed by the machine learning model to improve future financial fraud predictions by the machine learning model. 
 
     
     
       2. The method of  claim 1 , wherein the features include a community size feature indicating a number of customer accounts within a community. 
     
     
       3. The method of  claim 2 , wherein the features include a community structure feature indicating a structure of the community. 
     
     
       4. The method of  claim 3 , wherein the features include a suspicious activity report (SAR) feature indicating a number of customer accounts within the community that are associated with a SAR. 
     
     
       5. The method of  claim 4 , wherein the features include a transaction feature indicating financial transaction information associated with the customer accounts in the community. 
     
     
       6. The method of  claim 5 , wherein the machine learning model uses a plurality of factors in determining financial fraud risk including a total increase in a number of SARs associated with a community. 
     
     
       7. The method of  claim 6 , wherein the plurality of factors in determining financial fraud risk includes a rate of change in the number of SARs associated with a community. 
     
     
       8. The method of  claim 7 , wherein the plurality of factors in determining financial fraud risk including a rate of change in the structure of the community. 
     
     
       9. The method of  claim 8 , further comprising updating the machine learning model with new training data based on the determination of financial fraud risk performed using the machine learning model that have been proven to be accurate in identifying financial fraud. 
     
     
       10. A system configured to perform financial fraud assessment, the system comprising memory for storing instructions, and a processor configured to execute the instructions to:
 receive a training dataset that includes financial data that have been proven to be associated with financial fraud; 
 train a machine learning model based on an analysis of features from the training dataset to identify patterns in the financial data that have been proven to be associated with financial fraud; 
 receive financial information associated with customer accounts; 
 establish communities for the customer accounts; 
 create a baseline set of the features for each of the communities, wherein the baseline set of the features comprises a first total dollar amount of all transactions between all the customer accounts in the community; 
 receive new financial information associated with the customer accounts; 
 update the communities for the customer accounts based on the new financial information; 
 extract an updated set of the features for each of the communities, wherein the updated set of the features comprises a second total dollar amount of all transactions between all the customer accounts in the community; 
 determine a difference between the baseline set of the features and the updated set of the features for each of the communities; and 
 perform, by the machine learning model, a financial fraud risk assessment for each of the communities to identify one or more communities that have an increased risk of financial fraud by determining whether the difference between the baseline set of the features and the updated set of the features for each of the communities is indicative of an increased risk of fraud based on the patterns identified in the financial data that have been proven to be associated with financial fraud during training of the machine learning model, wherein there is an increased risk of financial fraud for the community when a difference in the second total dollar amount and the first total dollar amount for the community exceeds a threshold determined by the machine learning model that is indicative increased risk of financial fraud for the community; and 
 retrain the machine learning model based on an accuracy determination of the financial fraud risk assessment performed by the machine learning model to improve future financial fraud predictions by the machine learning model. 
 
     
     
       11. The system of  claim 10 , wherein the features include a community size feature indicating a number of customer accounts within a community. 
     
     
       12. The system of  claim 11 , wherein the features include a suspicious activity report (SAR) feature indicating a number of customer accounts within the community that are associated with a SAR. 
     
     
       13. The system of  claim 11 , wherein the features include a transaction feature indicating financial transaction information associated with the customer accounts in the community. 
     
     
       14. The system of  claim 13 , wherein the machine learning model uses a plurality of factors in determining financial fraud risk including a total increase in a number of suspicious activity reports (SARs) associated with a community. 
     
     
       15. The system of  claim 14 , wherein the plurality of factors in determining financial fraud risk includes a rate of change in the number of SARs associated with a community. 
     
     
       16. The system of  claim 15 , wherein the plurality of factors in determining financial fraud risk including a rate of change in a structure of the community. 
     
     
       17. A computer program product for performing financial fraud assessment, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
 receive a training dataset that includes financial data that have been proven to be associated with financial fraud; 
 train a machine learning model based on an analysis of features from the training dataset to identify patterns in the financial data that have been proven to be associated with financial fraud; 
 receive financial information associated with customer accounts; 
 establish communities for the customer accounts; 
 create a baseline set of the features for each of the communities, wherein the baseline set of the features comprises a first total dollar amount of all transactions between all the customer accounts in the community; 
 receive new financial information associated with the customer accounts; 
 update the communities for the customer accounts based on the new financial information; 
 extract an updated set of the features for each of the communities, wherein the updated set of the features comprises a second total dollar amount of all transactions between all the customer accounts in the community; and 
 determine a difference between the baseline set of the features and the updated set of the features for each of the communities; and 
 perform, by the machine learning model, a financial fraud risk assessment for each of the communities to identify one or more communities that have an increased risk of financial fraud by determining whether the difference between the baseline set of the features and the updated set of the features for each of the communities is indicative of an increased risk of fraud based on the patterns identified in the financial data that have been proven to be associated with financial fraud during training of the machine learning model, wherein there is an increased risk of financial fraud for the community when a difference in the second total dollar amount and the first total dollar amount for the community exceeds a threshold determined by the machine learning model that is indicative increased risk of financial fraud for the community; and 
 retrain the machine learning model based on an accuracy determination of the financial fraud risk assessment performed by the machine learning model to improve future financial fraud predictions by the machine learning model. 
 
     
     
       18. The computer program product of  claim 17 , wherein the features include a community size feature indicating a number of customer accounts within a community, a suspicious activity report (SAR) feature indicating a number of customer accounts within the community that are associated with a SAR, and a transaction feature indicating financial transaction information associated with the customer accounts in the community. 
     
     
       19. The computer program product of  claim 18 , wherein the machine learning model uses a plurality of factors in determining financial fraud risk including a total increase in a number of SARs associated with a community, a rate of change in the number of SARs associated with a community, and a rate of change in a structure of the community. 
     
     
       20. The computer program product of  claim 18 , wherein the machine learning model is updated with new training data based on the determination of financial fraud risk performed using the machine learning model that have been proven to be accurate in identifying financial fraud.